Soil organic matter (SOM) is essential to improve soil and ecosystem functioning. The concepts of soil quality and capability (Bouma et al., 2017; Bünemann et al., 2018) imply the capacity of a soil to sustain diverse functions, including biomass production, organic carbon (C) storage, and the promotion of biodiversity. From this perspective, SOM is a key component, because it is an important source of C and also of a variety of organic material (including nitrogen and metal-compounds) that can be used as energy and a nutrient source to support the soil microbiota and the consequent food chain. In this context, for example, low SOM can serve as a limitation, usually yielding lower biological diversity and biomass (plants, fauna, and microbiota) (Nielsen et al., 2011; Luo et al., 2018). On the other hand, soils with higher SOM usually have higher biodiversity, biomass and more diverse ecosystem functions (Xu et al., 2013; Bastida et al., 2016; Luo et al., 2018; Tian et al., 2018). Currently, SOM is receiving increasing attention due to the potential of soils to store carbon C and reduce atmospheric CO₂ levels, while at same time improving soil fertility (Lammel et al., 2017; Garratt et al., 2018).

Soil organic matter is not only important because of the role that it plays in soil carbon and nitrogen cycles, but also because SOM affects soils in three important ways: it acts as a substrate for microbial activity, as an agent to bind different mineral fractions in soil aggregates, and as an agent to change soil physical-chemical characteristics by improving mineral nutrient storage, water storage, and chelating pollutants (Cotrufo et al., 2015; Han et al., 2016; Zheng et al., 2016). Thus, SOM usually increases soil fertility and improves soil structure by increasing aggregation, porosity, and resistance to erosion (Zheng et al., 2016; Totsche et al., 2018). Consequently, SOM shapes habitats for soil biodiversity and improves plant growth, thus affecting the whole associated food web (Lammel et al., 2015; Laliberté et al., 2017; Baveye et al., 2018).

Soil carbon turnover and storage in soil are complex processes that involve microbial activity in complex physical matrices (Baveye et al., 2018; Totsche et al., 2018). Depending on the local physical-chemical characteristics of the soil, biological litter (that includes dead tissue from plants, animals, and microorganisms) is decomposed in the soil, stimulating the activity and diversity of biota, including bacteria, fungi, protists, and soil fauna, and yielding SOM (Cotrufo et al., 2015). In our article, we differentiate biological material as the organisms' tissues that can be morphologically recognized in the soil samples and SOM as the amorphous decomposed material that cannot be related to its biological origin anymore. During SOM turnover and storage in soil, the spatial distribution of C in the aggregates and pores is critical for biological activity and soil structure (Van Loo et al., 2014; Steffens et al., 2017; Tecon and Or, 2017; Totsche et al., 2018). On the other hand, the SOM position is associated with the attachment of mineral particles forming the aggregates and pores themselves (Totsche et al., 2018). This also includes plant residues and fungal hyphae, which may foster soil aggregation and thus pore properties (Oades and Waters, 1991). The pores vary in size from few nm to mm, and their size regulates water, solute and air diffusion rates. The pore size also compartmentalizes biotic habitats, as for example some bacterial cells can occupy pores with <1 μm, while fungal hyphae usually have diameters of 5–10 μm and thus need more space (Tecon and Or, 2017; Totsche et al., 2018). In addition, the spatial location of SOM will make it available or unavailable for microbial activity (Tecon and Or, 2017; Totsche et al., 2018).

To gain a better understanding of the interaction of SOM with biotic and abiotic factors, better analytical tools are needed, especially considering the spatial 3D structure of the soil matrices (Baveye et al., 2018). Among the diverse techniques that are currently used (for details see the review from Baveye et al., 2018), one of the more promising techniques entails micro-computerized X-ray tomography (μCT). It serves as a powerful tool to analyze soil structure, including aggregates and porosity (Zabler et al., 2008; Peth et al., 2014; Couradeau et al., 2018; Quigley et al., 2018; Totsche et al., 2018). The μCT technique may also be used as a non-destructive method to analyze the soil biota, for example, detecting fine roots and micro-fauna (Mooney et al., 2012; Guimarães et al., 2017). Thus, μCT is an important technique that can be coupled with other methods, such NanoSIM and SEM, to better elucidate the interface of physical, chemical, and biological aspects of the soil (Baveye et al., 2018; Schlüter et al., 2019). Previous studies have investigated the use of contrast agents to locate SOM and for qualitative analysis of biological material using μCT (Boyde et al., 2014; Peth et al., 2014; Van Loo et al., 2014; Li et al., 2015). The principle is that organic matter has strong affinity for some heavy elements and that these elements (with Z > 30) increase the X-ray attenuation, thus improving imaging contrast (Van Loo et al., 2014). Among the most effective agents for soil, silver (Ag) stands out (Van Loo et al., 2014; Niehaus et al., 2016). Several other heavy metals bind with SOM, such as Cu, Co, Fe, Mn, Mo, and Zn, but since they naturally occur in soil, they should be avoided as contrast agents to reduce confounding effects. Other efficient agents include lead (Pb) and osmium (Os) tetroxide (Peth et al., 2014; Rawlins et al., 2016), but these are more toxic materials whose use should be avoided when possible (Van Loo et al., 2014). Iodine (I) is another important staining agent for organic matter in soil and its gaseous form I₂ has been recently reported as an effective contrast for μCT of animal specimens (Boyde et al., 2014; Li et al., 2015). Eosin (Br based) was likewise reported as a highly effective enhancer for μCT of human tissues (Busse et al., 2018), but both gaseous I₂ and eosin, have not been tested in soil samples yet. In addition, the use of some heavy metals enables K-edge measurements. Every element has a K-edge energy value that excites the K-shell electron of an atom. When an atom is stimulated with that specific energy, there is an abrupt change in the X-ray attenuation value, thus allowing the detection of that specific element. Thus, K-edge μCT measurements allow the spatial imaging of specific elements, such as the contrast agents Ag and I, and thus improve the indirect detection of organic material in the soil samples (Peth et al., 2014; Egan et al., 2015).

New advances in μCT image resolution, coupled or not with K-edge measurements, enable greater accuracy to locate and image SOM and biological material in soil, such as plant material and fungal hyphae. The best resolution of μCT can be achieved by using synchrotron radiation (SR). However, access to SR-μCT is still restricted to a few facilities worldwide and thus measurement time is limited due to sharing the facility with other research fields. Since time and access to the facilities is limited, the report of new technical improvements is highly relevant and in turn can benefit other researchers in the environmental sciences. Likewise, this may contribute to a better understanding of the interaction between SOM, soil organisms, like fungi, the soil matrix and associated biological processes (Lehmann and Rillig, 2015; Tecon and Or, 2017; Totsche et al., 2018).

While standard μCT mainly allows the qualitative evaluation of biological material in the samples, such as morphological aspects of specimens (Mooney et al., 2012; Boyde et al., 2014; Busse et al., 2018), the K-edge SR-μCT can generate quantitative information about the SOM distribution (Peth et al., 2014; Rawlins et al., 2016). Qualitative improvement of the images by contrast agents may favor better root and soil organism detection by improving the attenuation density of the biological material (e.g., root) in relation to the soil pore space (Mooney et al., 2012). The K-edge SR-μCT quantitative approach is limited to target some elements with high Z that bind to the SOM (e.g., Ag or I₂), so the test of the applicability of such elements is the first step for future quantitative application of this approach.

The objective of this study was to test different contrast agents, including silver (Ag), eosin (Br based) as well as liquid and gaseous iodine (I), in order to qualitatively analyze SOM and biological material in soil samples using SR-μCT. The second objective was to test K-edge μCT for the Ag and I₂ treatments, and the third objective to test the I₂ staining using conventional μCT. To the best of our knowledge, eosin and gaseous I₂ have not previously been reported in tests for μCT of biological material in soil samples. Likewise, I and Ag were not previously reported as staining agents of SOM using K-edge SR-μCT of soil samples. Our expectation was that I₂ is the best agent. In its gaseous form, it potentially has higher penetration power in soil aggregates and produces less impact on the soil structure while other liquid contrast agents could be less efficient (Baveye et al., 2018).

In the first experiment, we tested six staining methods (liquid and gaseous contrast agents) to improve the imaging of SOM and biological material in soil samples. Of all the treatments, qualitatively, iodine returned the best images and yielded higher contrast for the biological material, including root fragments and hyphae-like structures (Figures 1, 2). The biological material in the samples treated with Lugol and I₂ had higher X-ray attenuation (are brighter in the image), when compared to the control sample and the sample treated with Eosin + Ag (Figure 1). Furthermore, few details can be observed in the particulate organic matter in the control (Figure 2, details 1. A-B, green arrows), while structures that look like microbial colonies are observed in the sample treated with iodine (Figure 2.2, Lugol/iodine, green arrow). Moreover, structures resembling fungal hyphae were observed in the sample treated with AgNO₃ (Figure 2.4, green arrow), but similar structures are much brighter in the sample treated with I₂ (Figure 2.6A, I2, green arrow). Additionally, we found that a partially decomposed root fragment was externally well-stained by iodine (very bright), because organic materials do not have increased X-ray attenuation effects (Figure 2.6B, I2, green arrow), and because that fragment was very rich in iodine, an element that increases the X-ray attenuation and consequently renders brighter images (Supplementary Figure 1.B1, K-edge colored image). In an additional attempt to differentiate the biological material, we tried to apply a machine learning segmentation tool to separate organic and mineral material, but it worked relatively well only for the Ag treatment (details in Supplementary Figure 2). In that image a structure resembling a fungal hypha is indicated by a green arrow (it is similar to the structure observed in Figure 2.4, green arrow).

The Ag and I₂ treatments were further evaluated by qualitative and quantitative measurements of the staining process. After the SR-μCT image acquisition (Figures 1, 2), the Ag and I₂ treatments were submitted to K-edge SR-μCT. Both elements were successfully detected in the samples and thus highlighted the stained SOM location in the samples (Figure 3 and Supplementary Figure 1). The decomposing root (Figure 2.6B, I2, green arrow) can be seen stained by I₂ in the 3D image (Figure 3.1, green arrow, and Supplementary Figure 1.B1). For I₂, we confirmed that the organic matter in the interior of the aggregates was stained (Supplementary Figure 1.B1). While the sand particles (S) were not stained, the interior of the root fragment (green arrow) and the interior of two soil aggregates (blue arrows) were intensely stained by I₂ (Supplementary Figure 1.B1). For the Ag treatment these differences are not evident, since overall the Ag staining was not effective (Supplementary Figure 1.B2).

We then estimated the quantity of the material bound to SOM by quantification of the volume related to the differential attenuation of measurements before and after the K-edges. For I₂, the stained sample volume is estimated between 3.0 and 4.7% of the full sample (including air); and considering a porosity estimated of 28%, we estimated that the I₂ staining was 0.8–1.3% of the soil sample (value close to the 1.1% C content of the sample). For Ag, we estimated that the staining was 0.1–0.3% of the matter inside the sample (much lower than the 1.1% C content).

In the second experiment, we used standard μCT to qualitatively evaluate the I₂ staining in a variety of samples (Figures 4, 5; Supplementary Figures 3, 4). There were no visual differences for the mineral fraction in the control treatment before and after the staining (Figures 4.1A–C). On the other hand, the root is clearly more visible after the I₂ incubation (Figures 4.1A–C, blue circles and Figures 5.1A,B, blue circles; there was a little movement of the root inside the tube during the 4 week incubation period). It is also clear in the image that the nylon mesh in the bottom of the tube, as the polyimide tube itself, was also stained by I₂ (Figures 5.1A,B). There was also improved detection of the soil micro-animals added to the tubes. The three Collembola individuals could not be detected at all before incubation with I₂ (Figures 4.1A–C), but they are very evident after the 4 week incubation in I₂ (Figures 5.2A,B).

We also analyzed soil cores (undisturbed samples) that were submitted to I₂ incubation (Figures 4.2A–C). For all the larger samples, longer incubation time was required. Organic material attenuation was improved, and material that was barely visible before incubation became visible (Figures 4.2A–C blue circles and Figures 5.4A,B and Supplementary Figure 3). Additionally, higher attenuation was also observed in parts of the individual aggregates samples, that we assume they are organic material clusters (Figures 5.3A,B and also in Supplementary Figure 4, highlighted by circles in the images).

The use of higher resolution techniques for spatial imaging of SOM and biological material in soil samples is crucial to identify how these components interact with soil structure and fertility, microbial activity and ultimately affecting ecosystem functioning (Cotrufo et al., 2015; Tecon and Or, 2017; Baveye et al., 2018; Garratt et al., 2018). Among several options (Vidal et al., 2018; Gorka et al., 2019), as reviewed in detail by Baveye et al. (2018), one promising technique is μCT and efforts have been made to improve it, such as using contrast agents and K-edge μCT (Peth et al., 2014; Van Loo et al., 2014). Here, we tested a variety of contrast agents and report that gaseous I₂ has a good potential to improve the spatial imaging of SOM and biological material in soil. Moreover, we report images with probably the best image resolution up to date for these techniques in soil samples.

In our study, iodine was not only a good contrast agent, but the use of this element, as also the Ag staining, allowed the use of K-edge μCT. While I₂ had a good staining efficiency (the K-edge measurement values are very similar to those obtained by the chemical analysis of C), the Ag stained had low efficiency. In previous studies, using conventional μCT, the use of Ag improved contrast of organic material in soil (Van Loo et al., 2014) and bacterial material in catheter biofilms (Niehaus et al., 2016). Ag has high affinity to bind to SOM, polyssaccharidic debris of plants and microbial cell walls, forming chelates and complexes with organic material, for example, containing carboxyl, phenolic, alcoholic and enolic-OH, and carbonyl structures of various types (Chenu and Plante, 2006; Van Loo et al., 2014). The use of iodine is also reported as an enhancer for CT imaging for animals samples, due to its high affinity to organic material (Boyde et al., 2014; Li et al., 2015). The four main mechanisms for iodine staining are as follows: (1) the reduction from I₂ to iodide and triiodide ions which may form bonds with cations; (2) the formation of complexes with starch, glycogen and glycoproteins; (3) the formation of organo-iodine compounds (after I₂ reacts with C-C double bonds in unsaturated molecules); and (4) the displacement of covalently bound chlorine atoms (Yamaguchi et al., 2010; Boyde et al., 2014). For animal specimens, Boyde et al. (2014) suggested that longer incubation with I₂, up to several weeks, could yield better staining results. In the short incubation period of experiment 1 (24 h), the root fragments were mainly externally stained by I₂ (Figure 2.6). So, likely longer incubation times could yield better internal staining (Boyde et al., 2014). However, future studies should still investigate the ability of I₂ to stain less porous organic material, such as plant remnants. Although, the contrast agents Ag and iodine have high affinity with SOM, fortunately their interaction with colloidal materials, like clays and negatively charged minerals, have been reported as negligible (Chenu and Plante, 2006; Yamaguchi et al., 2010; Van Loo et al., 2014). On the other hand, theoretically cationic Ag could also bind to the cation exchange capacity of clays, and this possibility should be better investigated in future studies. In our study, Ag was a poor agent and we did not detect any binding to inorganic material. We also did not observe binding of I₂ to inorganic material, but we recommend that after the incubation it is essential to keep the samples in open air to evaporate residual I₂ before the μCT analysis (see method).

We did not observe image improvement using eosin in our experiment. In a previous study, eosin stained soft human tissues well, since samples were soaked in a concentrated solution and later thoroughly dried, allowing for a high concentration of Br to accumulate in the samples (Busse et al., 2018). We surmise that since the organic material in our experiment was only a small fraction of the soil sample (usually 1–6%) and SOM has very distinct characteristics from soft organic tissues used in the previous study, we did not achieve the same results. Additionally, eosin is a more specific staining for biological cells, that binds mainly to cytoplasmic protein (Busse et al., 2018), while I₂ has a more general mechanism that attaches to various organic material (as described above). Moreover, we suggest that the liquid staining process may include chemicals (such as formaldehyde) and physical processing (e.g., centrifugation and dry-wet cycles), which could damage soil aggregates. So gaseous products, like I₂ or Os, look more adequate agents, since they potentially have higher penetration power and cause less disturbance into the samples (Boyde et al., 2014; Peth et al., 2014).

Furthermore, the K-edge μCT technique has proven to be a powerful tool for spatial location of elements in samples. Previous studies used gaseous Os to stain SOM in large soil macroaggregates (4–6 mm) and successfully located the SOM in the sample (Peth et al., 2014; Rawlins et al., 2016; Quigley et al., 2018). However, OsO₄ is highly toxic and the method was previously applied for large aggregates in a setup that can hardly be used for undisturbed soil samples. In the present study, we demonstrate that gaseous I₂ is a feasible alternative for the staining of soil samples and that it can be successfully applied to a mix of micro and macro-aggregates, with potential to be used also with undisturbed soil samples (Figure 5). Moreover, new technological advances may allow the use of K-edge μCT also with standard μCT machines when it is coupled with a spectral detector, thus facilitating easier access to this technique (Egan et al., 2015).

In our images, K-edge measurements with I₂ highly enhanced the location of SOM in the aggregates. It may allow future comparisons of soils with different physical-chemical properties and bring new insights into the drivers of associated microbial processes (Vidal et al., 2018; Gorka et al., 2019; Schlüter et al., 2019). For example, instead of correlating microbial diversity with total C and soil texture, future studies could correlate it with the SOM which is available at the surface of the aggregates and with the soil porosity analyzed by μCT (Tecon and Or, 2017; Baveye et al., 2018). Using μCT analysis may achieve better resolution to define the drivers of microbial diversity and activity. Improved imaging definition could likewise contribute to more precisely defining the role of SOM in binding mineral particles inside the aggregates, for example, estimating in contrasting soils the quantity of silt and clay aggregated by inorganic processes or by SOM. The review of Baveye et al. (2018) emphasizes the benefits of coupling different analytical tools, such as μCT, NanoSim, SEM, and also other techniques, allowing a better understanding of the interactions among physical, chemical, and biological aspects of the soil.

Lastly, the I₂ treatment qualitatively improved the visual definition of biological material in conventional μCT, including roots and soil animals (Collembola). Mooney et al. (2012) indicated that one of the challenges of μCT is to improve the attenuation density of the organic material in relation to the air pores. Thus, contrast agents can increase the attenuation of the organic material in relation to the air pore and consequently improving the image quality. The work by Mooney et al. (2012) was achieved on juvenile plants and often rather artificial soils. Indeed, the I₂ treatment would be useful to improve CT contrast of roots enabling their automated detection also in more complex and larger real life samples. The use of μCT may have a positive implication in soil biology, by improving a non-destructive method to detect fungi (SR-μCT), soil fauna (conventional μCT) and roots in soil samples. In the future, with better methodological improvements (e.g., better conventional μCT resolution), we suggest that methods that automatically count micro-fauna specimens in 2D images (Gilarranz et al., 2017) could be also improved and expanded for the 3D μCT images, as already occurs for roots (Mooney et al., 2012). And, as a side effect, the I₂ also stained the plastic containers of the samples (nylon and polyimide). So the I₂ staining procedure could have the potential to be applied also to the detection of plastic pollution in soil samples (Rillig et al., 2017).

Our results indicated that I₂ was the most effective contrast method for μCT of the soil samples and the use of the staining method coupled with K-edge μCT proved to be a powerful tool for spatial location of SOM. We acknowledge that the use of SR-μCT is an expensive and access-limited technique to study soil samples. The I₂ treatment also improved the visualization of biological material and consequently the qualitative analysis of fine plants roots and soil animals (Collembola). The use of μCT may have a positive implication in soil biology, by improving a non-destructive method to detect fungi (SR-μCT), soil fauna (conventional μCT), and roots in soil samples.

The datasets generated for this study are available on request to the corresponding author.

DL, TA, IM, and MR designed the study. DL and TA executed the experiments. TA processed the images. DL wrote and organized the article. All the authors contributed with the final text.